Analysis of data with strong 50Hz
=================================

by Chris Lidman, October 20, 2003


The 50Hz pickup and the horizontal stripes can be removed relatively
easily. Here is a simple algorithm that effectively removes the noise.

- Average (or median) the frame along detector rows and clip the highest
and lowest n pixels. A typical number is 250. Use a higher number if the
field is crowded or if the object extends over a substantial part of the
array.

- Subtract the resulting 1 dimensional image from the intial 2
dimensional image.

Generally speaking, one gets best results if this algorithm is applied
as the last step in the reduction process.

IRAF routine which does the above is attached. It was originally used
in processing SOFI data.

---------------------------------------------------------------------
# remove_lines
# CL script to remove bias lines from SOFI data
 
procedure remove_lines (in1)
 
        string in1   {prompt="List of input images"}
        int    nhigh {prompt="Number of high pixels"}
        int    nlow  {prompt="Number of low pixels"}
        

begin

        string tmpfile,tmpfile2,tmpfile3
        int    nh,nl

        nh  = nhigh
        nl  = nlow

          tmpfile = mktemp ("tmpia_tmp.")
          tmpfile2 = mktemp ("tmpia_tmp.")
          tmpfile3 = mktemp ("tmpia_tmp.")

          imtranspose(in1//"[-*,*]",tmpfile)
          imcom(tmpfile,tmpfile2,reject="minmax",
              project+,offset="",statsec="",zero="none",nhigh=nh,nlow=nl)
          imar(operand1=tmpfile,op="-",operand2=tmpfile2,result=tmpfile3)
          imtranspose(tmpfile3//"[*,-*]",in1//"jr")
                 
          imdelete (tmpfile, ver-, >& "dev$null")
          imdelete (tmpfile2, ver-, >& "dev$null")
          imdelete (tmpfile3, ver-, >& "dev$null")

end